Matches in SemOpenAlex for { <https://semopenalex.org/work/W2765084367> ?p ?o ?g. }
Showing items 1 to 77 of
77
with 100 items per page.
- W2765084367 abstract "In this paper, we present a novel approach to perform deep neural networks layer-wise weight initialization using Linear Discriminant Analysis (LDA). Typically, the weights of a deep neural network are initialized with: random values, greedy layer-wise pre-training (usually as Deep Belief Network or as auto-encoder) or by re-using the layers from another network (transfer learning). Hence, many training epochs are needed before meaningful weights are learned, or a rather similar dataset is required for seeding a fine-tuning of transfer learning. In this paper, we describe how to turn an LDA into either a neural layer or a classification layer. We analyze the initialization technique on historical documents. First, we show that an LDA-based initialization is quick and leads to a very stable initialization. Furthermore, for the task of layout analysis at pixel level, we investigate the effectiveness of LDA-based initialization and show that it outperforms state-of-the-art random weight initialization methods." @default.
- W2765084367 created "2017-11-10" @default.
- W2765084367 creator A5018014670 @default.
- W2765084367 creator A5066134818 @default.
- W2765084367 creator A5070046964 @default.
- W2765084367 creator A5073619925 @default.
- W2765084367 creator A5087037717 @default.
- W2765084367 date "2017-11-10" @default.
- W2765084367 modified "2023-10-16" @default.
- W2765084367 title "Historical Document Image Segmentation with LDA-Initialized Deep Neural Networks" @default.
- W2765084367 cites W2013066341 @default.
- W2765084367 cites W2058440010 @default.
- W2765084367 cites W2069683962 @default.
- W2765084367 cites W2076063813 @default.
- W2765084367 cites W2134262590 @default.
- W2765084367 cites W2136922672 @default.
- W2765084367 cites W2143888861 @default.
- W2765084367 cites W2149128132 @default.
- W2765084367 cites W2576852589 @default.
- W2765084367 cites W2773536003 @default.
- W2765084367 cites W4239943352 @default.
- W2765084367 doi "https://doi.org/10.1145/3151509.3151519" @default.
- W2765084367 hasPublicationYear "2017" @default.
- W2765084367 type Work @default.
- W2765084367 sameAs 2765084367 @default.
- W2765084367 citedByCount "14" @default.
- W2765084367 countsByYear W27650843672018 @default.
- W2765084367 countsByYear W27650843672019 @default.
- W2765084367 countsByYear W27650843672020 @default.
- W2765084367 countsByYear W27650843672021 @default.
- W2765084367 countsByYear W27650843672022 @default.
- W2765084367 countsByYear W27650843672023 @default.
- W2765084367 crossrefType "proceedings-article" @default.
- W2765084367 hasAuthorship W2765084367A5018014670 @default.
- W2765084367 hasAuthorship W2765084367A5066134818 @default.
- W2765084367 hasAuthorship W2765084367A5070046964 @default.
- W2765084367 hasAuthorship W2765084367A5073619925 @default.
- W2765084367 hasAuthorship W2765084367A5087037717 @default.
- W2765084367 hasBestOaLocation W27650843671 @default.
- W2765084367 hasConcept C108583219 @default.
- W2765084367 hasConcept C114466953 @default.
- W2765084367 hasConcept C124504099 @default.
- W2765084367 hasConcept C153180895 @default.
- W2765084367 hasConcept C154945302 @default.
- W2765084367 hasConcept C199360897 @default.
- W2765084367 hasConcept C41008148 @default.
- W2765084367 hasConcept C50644808 @default.
- W2765084367 hasConcept C69738355 @default.
- W2765084367 hasConcept C89600930 @default.
- W2765084367 hasConceptScore W2765084367C108583219 @default.
- W2765084367 hasConceptScore W2765084367C114466953 @default.
- W2765084367 hasConceptScore W2765084367C124504099 @default.
- W2765084367 hasConceptScore W2765084367C153180895 @default.
- W2765084367 hasConceptScore W2765084367C154945302 @default.
- W2765084367 hasConceptScore W2765084367C199360897 @default.
- W2765084367 hasConceptScore W2765084367C41008148 @default.
- W2765084367 hasConceptScore W2765084367C50644808 @default.
- W2765084367 hasConceptScore W2765084367C69738355 @default.
- W2765084367 hasConceptScore W2765084367C89600930 @default.
- W2765084367 hasLocation W27650843671 @default.
- W2765084367 hasLocation W27650843672 @default.
- W2765084367 hasOpenAccess W2765084367 @default.
- W2765084367 hasPrimaryLocation W27650843671 @default.
- W2765084367 hasRelatedWork W1563895814 @default.
- W2765084367 hasRelatedWork W1967559100 @default.
- W2765084367 hasRelatedWork W2897195263 @default.
- W2765084367 hasRelatedWork W2948658236 @default.
- W2765084367 hasRelatedWork W2960184797 @default.
- W2765084367 hasRelatedWork W2972093541 @default.
- W2765084367 hasRelatedWork W3115553566 @default.
- W2765084367 hasRelatedWork W3135174555 @default.
- W2765084367 hasRelatedWork W3209779739 @default.
- W2765084367 hasRelatedWork W4285827401 @default.
- W2765084367 isParatext "false" @default.
- W2765084367 isRetracted "false" @default.
- W2765084367 magId "2765084367" @default.
- W2765084367 workType "article" @default.